{
  "id": 94706,
  "title": "basic question about augmentation",
  "url": "/competitions/champs-scalar-coupling/discussion/94706",
  "author_name": "Jacek Poplawski",
  "post_date": "2019-06-06T10:46:21.838000",
  "votes": 8,
  "comment_count": 9,
  "views": 0,
  "content": "<p>Is scalar coupling between atom1 and atom2 same as between atom2 and atom1?</p>",
  "messages": [
    {
      "id": 546193,
      "postDate": "2019-06-06T10:46:21.837Z",
      "content": "<p>Is scalar coupling between atom1 and atom2 same as between atom2 and atom1?</p>",
      "rawMarkdown": "Is scalar coupling between atom1 and atom2 same as between atom2 and atom1?",
      "votes": 8
    },
    {
      "id": 546223,
      "postDate": "2019-06-06T11:17:15.403Z",
      "content": "<p>Yes.</p>\n\n<blockquote>\n  <p>The J coupling (always reported in Hz) isfield-independent (i.e. J is constant at different external magnetic field strength), and is mutual (i.e. JAX = JXA).</p>\n</blockquote>\n\n<p><a href=\"https://www.chem.wisc.edu/areas/reich/nmr/Notes-05-HMR-v26-part2.pdf\">J coupling</a></p>",
      "rawMarkdown": "Yes.\n\n&gt; The J coupling (always reported in Hz) isfield-independent (i.e. J is constant at different external magnetic field strength), and is mutual (i.e. JAX = JXA).\n\n[J coupling](https://www.chem.wisc.edu/areas/reich/nmr/Notes-05-HMR-v26-part2.pdf)\n",
      "votes": 3,
      "replies": [
        {
          "id": 546232,
          "postDate": "2019-06-06T11:28:15.223Z",
          "content": "<p>perfect, so we can insert the mirrored samples to train (=double train data size). will try this later..</p>",
          "rawMarkdown": "perfect, so we can insert the mirrored samples to train (=double train data size). will try this later..",
          "votes": 1
        },
        {
          "id": 546242,
          "postDate": "2019-06-06T11:38:33.853Z",
          "content": "<p>It will be interesting if you find something.I guess because we have a lot of data it won't give us something very useful.And it is also interesting to compare it with kfold!</p>",
          "rawMarkdown": "It will be interesting if you find something.I guess because we have a lot of data it won't give us something very useful.And it is also interesting to compare it with kfold!",
          "votes": 1
        },
        {
          "id": 546257,
          "postDate": "2019-06-06T11:52:20.353Z",
          "content": "<p>great, thanks!</p>",
          "rawMarkdown": "great, thanks!"
        },
        {
          "id": 546264,
          "postDate": "2019-06-06T12:05:38.613Z",
          "content": "<p>yea, I use 20% hold-out set for validation. much quicker and currently stable enough</p>",
          "rawMarkdown": "yea, I use 20% hold-out set for validation. much quicker and currently stable enough"
        },
        {
          "id": 547003,
          "postDate": "2019-06-07T05:55:03.450Z",
          "content": "<p>I tried double train size, so double train time, double ram usage. Features and model is exactly the same. MAE on valid improved: 0.40 -&gt; 0.38, lb also improved: -1.187 -&gt; -1.272</p>",
          "rawMarkdown": "I tried double train size, so double train time, double ram usage. Features and model is exactly the same. MAE on valid improved: 0.40 -&gt; 0.38, lb also improved: -1.187 -&gt; -1.272",
          "votes": 12
        },
        {
          "id": 547013,
          "postDate": "2019-06-07T06:30:21.770Z",
          "content": "<p>Ok! Thanks for the info!</p>",
          "rawMarkdown": "Ok! Thanks for the info!"
        },
        {
          "id": 563148,
          "postDate": "2019-06-28T00:44:05.943Z",
          "content": "<p>Thanks for the info! Did you only  change the atom_index?</p>",
          "rawMarkdown": "Thanks for the info! Did you only  change the atom_index?",
          "votes": 1
        }
      ]
    },
    {
      "id": 548011,
      "postDate": "2019-06-08T16:19:30.843Z",
      "rawMarkdown": "",
      "isDeleted": true
    }
  ],
  "comments": [
    {
      "id": 546223,
      "author_name": "George",
      "author_url": "",
      "post_date": "2019-06-06T11:17:15.403000",
      "content": "<p>Yes.</p>\n\n<blockquote>\n  <p>The J coupling (always reported in Hz) isfield-independent (i.e. J is constant at different external magnetic field strength), and is mutual (i.e. JAX = JXA).</p>\n</blockquote>\n\n<p><a href=\"https://www.chem.wisc.edu/areas/reich/nmr/Notes-05-HMR-v26-part2.pdf\">J coupling</a></p>",
      "votes": 3,
      "replies": [
        {
          "id": 546232,
          "author_name": "Michael Jahrer",
          "author_url": "",
          "post_date": "2019-06-06T11:28:15.223000",
          "content": "<p>perfect, so we can insert the mirrored samples to train (=double train data size). will try this later..</p>",
          "votes": 1,
          "replies": []
        },
        {
          "id": 546242,
          "author_name": "George",
          "author_url": "",
          "post_date": "2019-06-06T11:38:33.853000",
          "content": "<p>It will be interesting if you find something.I guess because we have a lot of data it won't give us something very useful.And it is also interesting to compare it with kfold!</p>",
          "votes": 1,
          "replies": []
        },
        {
          "id": 546257,
          "author_name": "Jacek Poplawski",
          "author_url": "",
          "post_date": "2019-06-06T11:52:20.353000",
          "content": "<p>great, thanks!</p>",
          "votes": 0,
          "replies": []
        },
        {
          "id": 546264,
          "author_name": "Michael Jahrer",
          "author_url": "",
          "post_date": "2019-06-06T12:05:38.613000",
          "content": "<p>yea, I use 20% hold-out set for validation. much quicker and currently stable enough</p>",
          "votes": 0,
          "replies": []
        },
        {
          "id": 547003,
          "author_name": "Michael Jahrer",
          "author_url": "",
          "post_date": "2019-06-07T05:55:03.450000",
          "content": "<p>I tried double train size, so double train time, double ram usage. Features and model is exactly the same. MAE on valid improved: 0.40 -&gt; 0.38, lb also improved: -1.187 -&gt; -1.272</p>",
          "votes": 12,
          "replies": []
        },
        {
          "id": 547013,
          "author_name": "George",
          "author_url": "",
          "post_date": "2019-06-07T06:30:21.770000",
          "content": "<p>Ok! Thanks for the info!</p>",
          "votes": 0,
          "replies": []
        },
        {
          "id": 563148,
          "author_name": "lichen",
          "author_url": "",
          "post_date": "2019-06-28T00:44:05.943000",
          "content": "<p>Thanks for the info! Did you only  change the atom_index?</p>",
          "votes": 1,
          "replies": []
        }
      ]
    },
    {
      "id": 548011,
      "author_name": "",
      "author_url": "",
      "post_date": "2019-06-08T16:19:30.843000",
      "content": "",
      "votes": 0,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "546193": "Is scalar coupling between atom1 and atom2 same as between atom2 and atom1?",
    "546223": "Yes.\n\n&gt; The J coupling (always reported in Hz) isfield-independent (i.e. J is constant at different external magnetic field strength), and is mutual (i.e. JAX = JXA).\n\n[J coupling](https://www.chem.wisc.edu/areas/reich/nmr/Notes-05-HMR-v26-part2.pdf)\n",
    "548011": ""
  }
}